Claim Missing Document
Check
Articles

Analisis Peramalan Harga Saham PT Unilever Indonesia Menggunakan Pemodelan LSTM dengan Optimasi PSO Dewinto Burhan; Isran K. Hasan; La Ode Nashar
Jurnal Riset Mahasiswa Matematika Vol 5, No 3 (2026): Jurnal Riset Mahasiswa Matematika
Publisher : Universitas Islam Negeri Maulana Malik Ibrahim Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.18860/jrmm.v5i3.40090

Abstract

Pergerakan harga saham cenderung tidak stabil dan sulit diprediksi karena memiliki pola yang kompleks serta berubah-ubah dari waktu ke waktu. Untuk mengatasi hal tersebut, penelitian ini menggunakan model Long Short-Term Memory (LSTM) dalam melakukan peramalan harga saham. Agar model yang dihasilkan memiliki kinerja yang optimal, dilakukan optimasi hyperparameter menggunakan metode Particle Swarm Optimization (PSO). Optimasi dilakukan pada tiga parameter utama LSTM yaitu LSTM units, dropout rate, dense units. Dari proses optimasi diperoleh enam konfigurasi model terbaik. Hasil pengujian menunjukkan model ke-5 memberikan hasil optimasi paling baik dengan nilai RMSE 120,3320 dan MAPE sebesar 3,53% dengan menggunakan kombinasi hyperparameter LSTM units = 137, dropout rate = 0,498, dan dense units = 32. Hasil ini menunjukkan bahwa optimasi hyperparameter memberikan peningkatan akurasi peramalan dibandingkan LSTM tanpa optimasi. Dengan demikian, kombinasi model LSTM dengan optimasi PSO mampu menghasilkan peramalan harga saham yang lebih akurat dan stabil. Pendekatan ini dapat digunakan sebagai alternatif dalam analisis pergerakan harga saham dan mendukung pengambilan keputusan investasi.
Prediksi Harga Emas Dunia Menggunakan Deep Learning GRU dengan Optimasi Nadam Ismail Saputra R. Harmain; Nurwan Nurwan; Isran K. Hasan; Djihad Wungguli; Nisky Imansyah Yahya
Jurnal Riset Mahasiswa Matematika Vol 4, No 6 (2025): Jurnal Riset Mahasiswa Matematika
Publisher : Universitas Islam Negeri Maulana Malik Ibrahim Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.18860/jrmm.v4i6.36007

Abstract

Volatilitas harga emas yang tinggi menuntut adanya metode prediksi yang andal untuk mendukung pengambilan keputusan investasi. Penelitian ini mengimplementasikan algoritma Gated Recurrent Unit (GRU) berbasis deep learning yang dioptimalkan menggunakan Nesterov-Accelerated Adaptive Moment Estimation (Nadam) untuk memprediksi harga emas harian.Model terbaik diperoleh dengan nilai Mean Squared Error (MSE) sebesar 0, 00012 pada data univariat dan 0, 00027 pada data multivariat. Mean Absolute Percentage Error (MAPE) yang diperoleh masing-masing sebesar 1,107% untuk data univariat dan 1,59% untuk data multivariat. Hasil tersebut mengindikasikan bahwa model GRU dengan optimasi Nadam memiliki performa prediksi yang tinggi, baik pada data deret waktu tanpa penambahan fitur maupun dengan penambahan fitur.
Perbandingan Model ARIMA-RBF dan ARIMA-GARCH dalam Peramalan Time Series Inflasi Provinsi Gorontalo Awalia Emiro; Isran K Hasan; Novianita Achmad
Research in the Mathematical and Natural Sciences Vol. 2 No. 1 (2023): November 2022-April 2023
Publisher : Scimadly Publishing

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55657/rmns.v2i1.76

Abstract

A quantitative method that is observed sequentially from time to time is a time series. In the real word, problems often occur where one method is not able to solve the problem. This research used linear and nonlinear methods by combining the ARIMA-RBF anda ARIMA-GARCH models in forecasting, and then the two models were compared based on the MAPE value. This research used monthly data on inflation for housing, water, electricity, and other fuels from 2008 to 2020. The forecast results from the ARIMA-RBF model obtained the MAPE value of 7.5%, and the ARIMA-GARCH model obtained the MAPE value of 11.8%. thus, the best model for predicting inflation in this research was the ARIMA RBF model.
Model Antrian Pelayanan Terhadap Nasabah Bank BRI Menggunakan Petri Net dan Aljabar Max Plus Sri Ayu Nurdin; Lailany Yahya; Isran K Hasan; Nurwan Nurwan
Research in the Mathematical and Natural Sciences Vol. 2 No. 2 (2023): May-October 2023
Publisher : Scimadly Publishing

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55657/rmns.v2i2.106

Abstract

Petri net is one model representing transitions and places connected by arrows. Max Plus Algebra is an algebraic structure in which all sets of real numbers  are equipped with max (maximum) and (addition). This research created a Petri net model of the customer service system for Bank BRI and a Max Plus Algebra model related to time to minimize service time at Bank BRI. The result is periodic time or characteristic values and vector characteristics where the values and are . The value of this vector's characteristics becomes a periodic time, which only takes 2 days 3 hours during working hours to disburse money after the client's arrival.
Peramalan Inflasi di Provinsi Gorontalo Menggunakan Metode General Regression Neural Network (GRNN) Isran K Hasan; Novianita Achmad; Putri Lapitung
Research in the Mathematical and Natural Sciences Vol. 3 No. 1 (2024): November 2023-April 2024
Publisher : Scimadly Publishing

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55657/rmns.v3i1.150

Abstract

Forecasting the inflation rate is important because the results obtained are used as an indicator that can influence the policies that will be made later. One policy that uses the results of this forecasting as one of the things that can influence it is economic policy and monetary policy. In this study, the method used is the general regression neural network (GRNN). This forecast is applied to inflation data in Gorontalo Province from January 2008 to April 2023, with the conclusion that it produces an inflation forecast for May – December 2023 with a MAPE value of 3.24% or an accuracy rate of 96.76%.